The FSM is in STANDBY state when arriving in vfio_ccw_mdev_probe(),
and this routine converts it to IDLE as part of its processing.
The error exit sets it to IDLE (again) but clears the private->mdev
pointer.
The FSM should of course be managing the state itself, but the
correct thing for vfio_ccw_mdev_probe() to do would be to put
the state back the way it found it.
The corresponding check of private->mdev in vfio_ccw_sch_io_todo()
can be removed, since the distinction is unnecessary at this point.
Fixes: 3bf1311f351ef ("vfio/ccw: Convert to use vfio_register_emulated_iommu_dev()")
Signed-off-by: Eric Farman <farman@linux.ibm.com>
Reviewed-by: Jason Gunthorpe <jgg@nvidia.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Link: https://lore.kernel.org/r/20220707135737.720765-3-farman@linux.ibm.com
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
        /*
         * Reset to IDLE only if processing of a channel program
         * has finished. Do not overwrite a possible processing
-        * state if the final interrupt was for HSCH or CSCH.
+        * state if the interrupt was unsolicited, or if the final
+        * interrupt was for HSCH or CSCH.
         */
-       if (private->mdev && cp_is_finished)
+       if (cp_is_finished)
                private->state = VFIO_CCW_STATE_IDLE;
 
        if (private->io_trigger)
 
        vfio_uninit_group_dev(&private->vdev);
        atomic_inc(&private->avail);
        private->mdev = NULL;
-       private->state = VFIO_CCW_STATE_IDLE;
+       private->state = VFIO_CCW_STATE_STANDBY;
        return ret;
 }